json-validator/include/jvalidate/detail/string.h

#pragma once

#if __has_include(<unicode/std_string.h>)
#define JVALIDATE_HAS_ICU
#include <unicode/brkiter.h>
#include <unicode/unistr.h>
#endif

namespace jvalidate::detail {
inline size_t length(std::string_view arg) {
#ifdef JVALIDATE_HAS_ICU
  icu::UnicodeString ucs = icu::UnicodeString::fromUTF8(icu::StringPiece(arg));
  return ucs.countChar32();
#else
  return arg.size();
#endif
}

inline std::string regex_escape(std::string_view arg) {
#ifdef JVALIDATE_HAS_ICU
  icu::UnicodeString const ucs = icu::UnicodeString::fromUTF8(icu::StringPiece(arg));
  if (ucs.countChar32() == arg.size()) {
    return std::string(arg);
  }

  UErrorCode status = U_ZERO_ERROR;
  std::unique_ptr<icu::BreakIterator> iter(
      icu::BreakIterator::createCharacterInstance(NULL, status));

  if (U_FAILURE(status)) {
    return std::string(arg);
  }

  icu::UnicodeString rval;
  iter->setText(ucs);
  int32_t start = iter->first();
  int32_t end = iter->next();
  while (end != icu::BreakIterator::DONE) {
    if (std::strchr("?*+", ucs.charAt(end))) {
      rval.append('(');
      rval.append(ucs, start, end - 1);
      rval.append(')');
      rval.append(ucs.char32At(end));
      end = iter->next();
    } else {
      rval.append(ucs, start, end - 1);
    }
    start = end;
    end = iter->next();
  }

  std::string out;
  return rval.toUTF8String(out);
#else
  return std::string(arg);
#endif
}
}